Integrating participatory science with official programmes using Bayesian machine learning to estimate beach macroplastic pollution in Spain

The integration of participatory science (PS) data into official monitoring frameworks offers a promising pathway to enhance the spatial and temporal coverage of environmental assessments. Significant efforts have been made within the framework of the Spanish National Marine Strategy, which transposes the Marine Strategy Framework Directive (56/2008/EC), to integrate citizen science data, particularly regarding the impacts of macroplastics. In this study, we analyze the methodological challenges and potential efficiencies of integrating official monitoring programme data on marine litter on beaches with participatory science data in Spain using Bayesian machine learning.

Leveraging a flexible Gaussian Process Regression framework, we model the spatial distribution of beach litter pollution along the Spanish coastline, accounting for the differing uncertainties inherent to the two data sources. This data-driven approach enables us to produce robust estimations of macroplastic pollution levels with associated uncertainty maps and identify locations where PS contributions significantly reduce the uncertainty of official monitoring efforts. Preliminary results include spatial predictions of marine beach litter density, uncertainty quantification along Spanish coastlines, and insights into the added value of PS data for underrepresented regions.

Beyond providing actionable insights for Spain, this study presents a globally adaptable blueprint for the assimilation of participatory science data into official environmental monitoring programmes. The present study demonstrates the potential of combining machine learning, official monitoring programmes and participatory science to achieve actionable science, with the aim of strengthening policy, optimising resource allocation and enhancing coastal management practices on a global scale.